Fix module loads

[becker33]

Fixes #5455

All methods within setup_package use an EnvironmentModifications object to control the environment. Those modifications are applied at the end of setup_package. Module loads for the build environment need to be done after the rest of the environment modifications are applied, as otherwise Spack will unset variables set by those modules.

This is a quick fix, but I don't see a more elegant solution to this problem.

[scheibelp]

IMO this should explicitly refer to the call to spack_env.apply_modifications and it may be work adding to the explanation like:

spack_env includes modifications like unsetting LD_LIBRARY path, which required modules may update. Modules are therefore loaded after the call to spack_env.apply_modifications.

[becker33]

@scheibelp I've updated the comment as you asked

[pramodskumbhar]

@becker33 : I have tested this on our x86 cluster and it's working fine. BUT, with this change, our Jenkins build is failing for TAU profiler build with error like:

     363   gcc -c -fPIC  -O2 -Wl,-rpath,/jenkins/01/workspace/hpc.spack.github/platform/cscsviz/opt/spack/linux-rhel6-x86_64/intel-17.0.0/tau-2.26.3-odsr7ym4v3uvfiartkswa7leqgi3quz6/lib -Wl,-rpath,/jenkins/01/workspace/hpc.spack.github/platform/cscsviz/opt/spack/linux-rhel6-x86_64/intel-17.0.0/tau-2.26.3-odsr7ym4v3uvfiartkswa7leqgi3quz6/lib64 -Wl,-rpath,/jenkins/01/workspace/hpc.spack.github/platform/cscsviz/opt/spack/linux-rhel6-x86_64/intel-17.0.0/pdt-3.24-itnh5oukvpbgmrwyilmfnnft4mdb65n5/lib -Wl,-rpath,/gpfs/bbp.cscs.ch/apps/viz/intel/2017/compilers_and_libraries_2017.0.098/linux/mpi/intel64/lib TauLinuxTimers.c
     367   In file included from TauLinuxTimers.c:5:
  >> 368   /usr/include/stdio.h:34:21: error: stddef.h: No such file or directory
     369   In file included from /usr/include/stdio.h:75,
     370                    from TauLinuxTimers.c:5:
  >> 371   /usr/include/libio.h:53:21: error: stdarg.h: No such file or directory
     372   In file included from /usr/include/stdio.h:75,
     373                    from TauLinuxTimers.c:5:
  >> 374   /usr/include/libio.h:332: error: expected specifier-qualifier-list before ‘size_t’
  >> 375   /usr/include/libio.h:364: error: expected declaration specifiers or ‘...’ before ‘size_t’
  >> 376   /usr/include/libio.h:373: error: expected declaration specifiers or ‘...’ before ‘size_t’
  >> 377   /usr/include/libio.h:491: error: expected declaration specifiers or ‘...’ before ‘__gnuc_va_list’
  >> 378   /usr/include/libio.h:493: error: expected declaration specifiers or ‘...’ before ‘__gnuc_va_list’
  >> 379   /usr/include/libio.h:495: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘_IO_sgetn’
     380   In file included from TauLinuxTimers.c:5:
  >> 381   /usr/include/stdio.h:80: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘va_list’
     382   In file included from TauLinuxTimers.c:5:

In above case, I am building tau with intel (or XL on bg-q) compiler but TAU does use GCC compiler to build some tools. Above error happens where those files get compiled with GCC. Its strange that above error is related to not finding standards headers from /usr/include.

Unfortunately I am not able to reproduce this in interactive session on our cluster. Do you have any idea what might be going wrong after this PR?

[scheibelp]

@pramodskumbhar could you post the full compiler entry? I was going to open a new issue for this but I think that detail would be useful in a better description for the issue.

[pramodskumbhar]

@scheibelp : I was debugging this and it seems like it was our Jenkins CI glitch. Now I am able to build everything fine. I will open an issue if I see this again.

Sorry for noise!